datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
codeforces-cots
Dataset Card for CodeForces-CoTs
Dataset description
CodeForces-CoTs is a large-scale dataset for training reasoning models on competitive programming tasks. It consists of 10k CodeForces problems with up to five reasoning traces generated by DeepSeek R1. We did not filter the traces for correctness, but found that around 84% of the Python ones pass the public tests.
The dataset consists of several subsets:
solutions: we prompt R1 to solve the problem and produce code.… See the full description on the dataset page: https://huggingface.co/datasets/open-r1/codeforces-cots.mmlu-pro-self-cot-deepseek-r1Use deepseek-r1 to generate COT in few-shot examples.
Aesir-Character-CoT-roleplay
Overview
Think with your role.
Most reasoning datasets teach models to think like an AI. This one teaches them to think like the character.
Continue updating until money run out, I will try to update this dataset in near future
Stats
1,973 high-quality conversations (filtered from 2,000 distilled — 27 dropped: prohibited content + missing-review + empty-content)
~14,349 assistant turns, each with full character-POV reasoning
Teacher: deepseek-v4-pro… See the full description on the dataset page: https://huggingface.co/datasets/beyoru/Aesir-Character-CoT-roleplay.cftc-cot
Chainticks CFTC COT
Normalized CFTC Commitments of Traders legacy futures rows from public-domain CFTC archives.
import pandas as pd
DATE = "YYYY-MM-DD"
URL = "https://huggingface.co/datasets/Chainticks/cftc-cot/resolve/main/legacy_futures/date={DATE}/part-0000.parquet"
df = pd.read_parquet(URL)
print(df.head())
Layout
legacy_futures/date=YYYY-MM-DD/part-0000.parquet
_schema.json
_manifest.json
LATEST_DATE.txt
Provenance
Rows must have… See the full description on the dataset page: https://huggingface.co/datasets/Chainticks/cftc-cot.China-K12-STEM-10K-CoT-Reasoning
K12-STEM-CoT-Chinese
1.54M Chinese K12 STEM problems with chain-of-thought solutions, 48% with diagrams.
The largest structured Chinese math/physics/chemistry reasoning dataset.
This is a curated sample (10,000 problems) of the full 1.54M dataset available via API.
Full Dataset Access
Access the full 1,540,000+ problems via API →
This Sample
Full API
Total problems
10,025
1,540,000+
With CoT solutions
10,025
1,490,000+
With diagrams
6,093
740,000+… See the full description on the dataset page: https://huggingface.co/datasets/lfaviate/China-K12-STEM-10K-CoT-Reasoning.OpenSakura-DS-260220-LN-ja-zh-COT-Lilith
OpenSakura Lilith LN COT Dataset
OpenSakura-DS-260220-LN-ja-zh-COT-Lilith is the COT/segment-level derivative built from the same LN source stream, with reasoning_content preserved.
Stats below are computed from the actual generated parquet files.
Dataset Summary
Metric
Value
Dataset ID
OpenSakura/OpenSakura-DS-260220-LN-ja-zh-COT-Lilith
Total rows
692,587
Total parquet files
233 (train: 162, arena: 12, reserve: 12, validation: 24, test: 23)
Total size
8… See the full description on the dataset page: https://huggingface.co/datasets/OpenSakura/OpenSakura-DS-260220-LN-ja-zh-COT-Lilith.gen3_cotrainingThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"action": {
"dtype": "float32",
"names": [
"right_joint_1.pos",
"right_joint_2.pos",
"right_joint_3.pos",
"right_joint_4.pos",
"right_joint_5.pos",
"right_joint_6.pos",
"right_joint_7.pos"… See the full description on the dataset page: https://huggingface.co/datasets/videron/gen3_cotraining.mmlu-pro-CoTnumina-math-llama-3.1-8b-bon-meta-cotDomofon-Cot-Conversations-700k
Domofon-Cot-Conversations-700k
Synthetic XML conversation data for training small language models on reasoning,
instruction following, XML formatting, and tool-use traces.
Repository: domofon/Domofon-Cot-Conversations-700k
What is inside
The dataset contains cleaned generated XML conversations from six families:
conv: multi-turn factual conversations with tool-use traces.
instruct: text-processing instructions, including deterministic count tool calls.
ds:… See the full description on the dataset page: https://huggingface.co/datasets/domofon/Domofon-Cot-Conversations-700k.gpqa_0shot_cotOpenThoughts3-456k-no-cotFable-5-CoT-TracesPersonal collection of Fable 5 reasoning traces.
Filter out the decoy ones and you're good.
Have fun! (Also, star my repo https://github.com/FusionCube18712/claude-codex-auto-resume if you can)
Happy distilling.
qa_verify_cot_new_6M_unfiltered_v7dataset_names = [
"HayatoHongoEveryonesAI/qa_verify_1m_cot_1",
"HayatoHongoEveryonesAI/qa_verify_1m_cot_2",
"HayatoHongoEveryonesAI/qa_verify_1m_cot_3",
"HayatoHongoEveryonesAI/qa_verify_1m_cot_4",
"HayatoHongoEveryonesAI/qa_verify_1m_cot_5",
"HayatoHongoEveryonesAI/qa_verify_2m_cot_2",
"HayatoHongoEveryonesAI/qa_verify_2m_cot_3",
]
https://colab.research.google.com/drive/1272DRwGt02zokQiHHOl4HpoKezdyw59O?usp=sharing
cftc-cot-weekly
CFTC Commitments of Traders weekly panel
Every CFTC Commitments of Traders report family in one tidy, model ready weekly
panel: harmonized positions, net positioning and COT index features, a market
reference map, and documented release provenance, from 1986 to last Friday, in
Parquet and CSV.
Dataset structure
Four tables, each its own named config (different schemas, never concatenated):
cot_panel_long: the tidy long panel, one row per (report_date, contract… See the full description on the dataset page: https://huggingface.co/datasets/Arimancy/cftc-cot-weekly.qa_verify_2m_cot_3qa_verify_cot_new_5.1M_v7HayatoHongoEveryonesAI/qa_verify_cot_new_6M_unfiltered_v7
https://colab.research.google.com/drive/1tjJ14xLa0UZ0slYqnRuQR8ngk1sPyY8v?usp=sharing
OpenThoughts3-456k-no-cot-with-olmo-system-promptOpenThoughts3-456k-gpt4.1-cotgpqa_0shot_cotchess-reasoning-cot-evalsGLM-5.2-CoT-Library
GLM-5.2 — CoT Library
A maintained mirror of publicly-available GLM-5.2 chain-of-thought datasets on Hugging Face — content-verified, deduplicated, and attributed to their original authors.
Dataset Viewer | Parquet
// what this is
A maintained library — a community mirror of publicly-available GLM-5.2 CoT datasets, aggregated, validity-filtered and content-verified, with per-row source attribution in first_source_dataset. It is not Crownelius' own data — every… See the full description on the dataset page: https://huggingface.co/datasets/Crownelius/GLM-5.2-CoT-Library.Task_Picupgloves_50fps_cotrain1This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"trossen_subversion": "v1.0",
"robot_type": "trossen_ai_mobile",
"total_episodes": 51,
"total_frames": 16676,
"total_tasks": 1,
"total_videos": 153,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 50,
"splits": {
"train": "0:51"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet"… See the full description on the dataset page: https://huggingface.co/datasets/Qipei/Task_Picupgloves_50fps_cotrain1.mmlu_pro_cotCoT-Collection-standardized
Dataset Card for "CoT-Collection-standardized"
More Information needed
Magpie-Reasoning-V2-250K-CoT-Llama3
Project Web: https://magpie-align.github.io/
Arxiv Technical Report: https://arxiv.org/abs/2406.08464
Codes: https://github.com/magpie-align/magpie
Abstract
Click Here
High-quality instruction data is critical for aligning large language models (LLMs). Although some models, such as Llama-3-Instruct, have open weights, their alignment data remain private, which hinders the democratization of AI. High human labor costs and a limited, predefined scope for prompting prevent… See the full description on the dataset page: https://huggingface.co/datasets/Magpie-Align/Magpie-Reasoning-V2-250K-CoT-Llama3.cot-hidden-state-trajectories
CoT Hidden-State Trajectories
Chain-of-thought traces and generation-time hidden-state activations from
11 open-weight language models, on Codeforces (competitive programming),
Hendrycks MATH, and SATBench (Boolean satisfiability).
This dataset accompanies the paper Reasoning Models Don't Just Think
Longer, They Move Differently (arXiv:2605.15454).
The paper asks whether reasoning-trained models follow different
hidden-state paths than matched instruction-tuned baselines, after… See the full description on the dataset page: https://huggingface.co/datasets/gjoelbye/cot-hidden-state-trajectories.AIME24-25_CoT_Verification
Dataset for ICLR 2026 Paper: Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs
📌 Dataset Summary
This dataset contains the rollouts (reasoning traces) and verification results used in our ICLR 2026 paper. The data allows for the analysis of how Reinforcement Learning with Verifiable Rewards (RLVR) incentivizes the correct reasoning of Large Language Models (LLMs) on challenging mathematics benchmarks.
The dataset… See the full description on the dataset page: https://huggingface.co/datasets/XumengWen/AIME24-25_CoT_Verification.Kimi-K3-CoT-Library
Kimi K3 — CoT Library
A maintained mirror of publicly-available Kimi K3 chain-of-thought datasets on Hugging Face — content-verified, deduplicated, and attributed to their original authors.
Dataset Viewer | Parquet
// what this is
A maintained library — a community mirror of publicly-available Kimi K3 CoT datasets, aggregated, validity-filtered and content-verified, with per-row source attribution in first_source_dataset. It is not Crownelius' own data — every… See the full description on the dataset page: https://huggingface.co/datasets/Crownelius/Kimi-K3-CoT-Library.cot-gemma4-26b-a4b
Gemma-4-26B-A4B-it Chain-of-Thought Oracle Corpus
Chain-of-thought rollouts generated with google/gemma-4-26B-A4B-it (MoE,
25.2B total / 3.8B active), in its native thinking mode, across a diverse suite
of reasoning tasks. Structure follows
ceselder/cot-oracle-corpus-v5
(CoT-only subset of the columns), built for chain-of-thought monitoring /
activation-oracle research.
2,121,354 rollouts over 212,161 unique problems (10 sampled
thinking rollouts per problem, temperature 0.8).… See the full description on the dataset page: https://huggingface.co/datasets/cds-jb/cot-gemma4-26b-a4b.